Joint Sequence-Structure Protein Design With Autoregressive Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing protein design methods struggle to efficiently generate both the amino acid sequence and structure of proteins in a coordinated manner, often requiring separate modeling processes that are computationally intensive and resource-heavy.
Innovation Solution
A protein design system utilizing a neural network architecture that jointly generates amino acid sequence and structure through an autoregressive process, incorporating an encoder, amino acid, and structure neural networks to create a protein design neural network that incrementally constructs the sequence and structure, allowing for more efficient and stable protein design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If separate modeling processes are used to generate amino acid sequence and structure, then each component can be optimized independently, but the overall computational resources and time required increase significantly
Solution Approach 1:
The patent combines separate sequence modeling and structure modeling into a unified autoregressive model that generates both amino acid sequences and their corresponding structures simultaneously. This integration allows the model to learn the joint distribution of sequence-structure pairs, improving coordination between sequence design and structural outcomes while reducing overall computational overhead through a single unified training process.
2Productivity
If autoregressive generation is used to incrementally construct sequence and structure, then computational resources are reduced, but the model complexity increases
Solution Approach 1:
The autoregressive model is segmented into distinct processing stages: encoding the input sequence, predicting structure parameters position-by-position, and generating outputs in an incremental manner. This segmentation allows the complex task of joint sequence-structure generation to be broken down into manageable sequential steps, reducing memory requirements and enabling efficient training while maintaining model sophistication.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for designing a protein by jointly generating an amino acid sequence and a structure of the protein. In one aspect, a method comprises: generating data defining the amino acid sequence and the structure of the protein using a protein design neural network, comprising, for a plurality of positions in the amino acid sequence: receiving the current representation of the protein as of the current position: processing the current representation of the protein using the protein design neural network to generate design data for the current position that comprises: (i) data identifying an amino acid at the current position, and (ii) a set of structure parameters for the current position; and updating the current representation of the protein using the design data for the current position.